PET/MR is a hybrid imaging modality that combines the exquisite soft tissue contrast of MR with the molecular information of PET. In order to utilize PET/MR in a clinical trial setting, images must be quantitatively accurate, and be reproducible across vendor platforms and institutions. Accurate MR-based attenuation correction (MR- AC) is currently a technical barrier to accomplishing these goals. A specific challenge is differentiating bone from air. While these tissue types have dramatic differences in the degree to which they attenuate photons, they both have negligible signal with conventional MR pulse sequences. Consequently, current MR-AC methods exhibit SUV errors of 20% or greater, particularly in areas within and adjacent to bone, and therefore, current PET/MR scanners do not meet the SUV accuracy required by NCI/ACRIN for clinical trials qualification. Ultra-short echo time (UTE) MR can capture signal in bone prior to its rapid signal decay and is a promising approach to achieve more accurate MR-AC. However, current UTE approaches have low image quality, clinically impractical acquisition times, and a field of view that is too limited for whole-body imaging. The goal of this academic-industrial collaboration is to address these current limitations of UTE by developing accurate and clinically practical methods for whole-body MR-AC, further refining novel and patented methods developed by our working group.
The specific aims to realize the goal: 1) Develop novel MR acquisition methods that maximize tissue information regarding photon attenuation for whole-body imaging. We will use our preliminary work in brain as a starting point, which employs an undersampled UTE-Dixon acquisition. 2) Establish image processing methods for determining photon attenuation on a voxel-level. Pattern recognition methods will be developed to analyze the combination of features extracted from the UTE- Dixon data sets. The photon attenuation will be estimated on a continuous scale reflecting the fractional composition of different tissue types within each voxel and also by directly mapping to CT values. 3) Demonstrate clinical feasibility of the above proposed MR-based attenuation correction methods. Clinical scanning with a commercial PET/MR system will be performed in a cancer patient population comparing the developed MR-AC methods to CT-AC values for SUV accuracy, image quality, and diagnostic accuracy. By bringing together cutting-edge advances in both MR acquisition and image analyses, the successful completion of these aims will achieve SUVs that are within 5% of those obtained with PET/CT (reference standard) with clinically appropriate acquisition time, image quality, and diagnostic accuracy, capable of supporting quantitative clinical trials with commercial PET/MR systems.

Public Health Relevance

PET/MR, a new imaging method, combines the soft-tissue detail and functional information of MRI with the molecular data gathered from PET. A current limitation of the new technology is that MR-derived correction for attenuation is rudimentary making in vivo metabolism measurements inaccurate. We will develop and demonstrate new methods that achieve accuracy specifications for PET/MRI systems that support quantitative measurements for clinical trials as well as serial monitoring of treatment response. These proposed improvements will not only impact cancer patients enrolling in clinical trials, but may also positively impact other populations such as pediatric and dementia patients.

National Institute of Health (NIH)
National Cancer Institute (NCI)
Research Project (R01)
Project #
Application #
Study Section
Special Emphasis Panel (ZRG1-SBIB-D (57))
Program Officer
Zhang, Huiming
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
Case Western Reserve University
Schools of Medicine
United States
Zip Code
Qian, Pengjiang; Xi, Chen; Xu, Min et al. (2018) SSC-EKE: Semi-Supervised Classification with Extensive Knowledge Exploitation. Inf Sci (N Y) 422:51-76
Qian, Pengjiang; Jiang, Yizhang; Wang, Shitong et al. (2017) Affinity and Penalty Jointly Constrained Spectral Clustering With All-Compatibility, Flexibility, and Robustness. IEEE Trans Neural Netw Learn Syst 28:1123-1138
Qian, Pengjiang; Zhao, Kaifa; Jiang, Yizhang et al. (2017) Knowledge-leveraged transfer fuzzy C-Means for texture image segmentation with self-adaptive cluster prototype matching. Knowl Based Syst 130:33-50
Qian, Pengjiang; Jiang, Yizhang; Deng, Zhaohong et al. (2016) Cluster Prototypes and Fuzzy Memberships Jointly Leveraged Cross-Domain Maximum Entropy Clustering. IEEE Trans Cybern 46:181-93
Qian, Pengjiang; Sun, Shouwei; Jiang, Yizhang et al. (2016) Cross-domain, soft-partition clustering with diversity measure and knowledge reference. Pattern Recognit 50:155-177
Su, Kuan-Hao; Hu, Lingzhi; Stehning, Christian et al. (2015) Generation of brain pseudo-CTs using an undersampled, single-acquisition UTE-mDixon pulse sequence and unsupervised clustering. Med Phys 42:4974-86
Hu, Lingzhi; Su, Kuan-Hao; Pereira, Gisele C et al. (2014) k-space sampling optimization for ultrashort TE imaging of cortical bone: applications in radiation therapy planning and MR-based PET attenuation correction. Med Phys 41:102301